Optimization of a Microscale, Axial-Flux, Permanent-Magnet Generator

D. P. Arnold1*, I. Zana1†, F. Herrault1, P. Galle1, J.-W. Park1, S. Das2, J. H. Lang2, and M. G. Allen1
1
School of Electrical and Computer Engineering, Georgia Institute of Technology, Atlanta, GA 30332, USA
Phone: (404) 894-9419, Fax: (404) 894-5028, Email: mark.allen@ece.gatech.edu
2
Dept. of Electrical Eng. and Computer Science, Massachusetts Institute of Technology, Cambridge, MA, 01239, USA
*
now with Dept. of Electrical and Computer Engineering, University of Florida, Gainesville, FL, 32611, USA
†
now with Center for Materials for Information Technology, University of Alabama, Tuscaloosa, AL, 35487, USA

Abstract
This paper presents the design optimization and characterization of a microscale, permanent-magnet (PM) generator, capable
of supplying 8 W of DC power to a resistive load at a rotational speed of 305,000 rpm. The generator is a three-phase, axial-
flux, PM machine, consisting of a stator with Cu surface windings and a multi-pole SmCo PM rotor. Optimization of the
machine geometries have resulted in a 30% improvement in power density over a previously reported machine (at 120,000
rpm). Furthermore, these design improvements in combination with higher rotational speeds has enabled a >7x improvement
in total output power and a net power density of 59 W/cm3.
Keywords: power generator, magnetic generator, permanent-magnet machine, FEMLAB finite element modeling

1. INTRODUCTION tradeoffs required to maximize the output power through

Microscale, axial-flux, PM generators are an active area of optimization of the overall machine geometry and stator
research for high power density electromechanical energy winding patterns.
conversion. These devices are of particular interest because
2. MACHINE OVERVIEW
their planar geometry merges well with MEMS fabrication
The generator is a three-phase, eight-pole, axial-flux,
techniques. These electrical power converters require input
synchronous machine comprised of a rotor with an annular
mechanical power, and hence rely on power from an
PM and soft magnetic back iron and a stator with Cu
integrated turbine or some other mechanical power source,
surface windings on a magnetically soft substrate, which
such as a gas-fueled turbine engine [1].
serves as a magnetic back iron.
We have previously reported a first-generation microscale,
The stator uses interleaved, electroplated Cu windings that
axial-flux, PM generator that demonstrated 2.6 W of
are dielectrically isolated from a 1-mm thick NiFeMo
mechanical-to-electrical power conversion and delivery of
substrate by a 5-µm polyimide layer (Fig.1). The rotor
1.1 W of DC power to a resistive load at a rotor speed of
contains an eight-pole, 500 µm thick, annular SmCo PM
120 krpm using a 9.5 mm diameter SmCo rotor [2]. We
and a 500 µm thick FeCoV (Hiperco 50) back iron. For this
have also reported methods to increase the rotational speeds
design, both the rotor magnet and back iron have an inner
and thus boost the output power by improving the rotor
diameter of 5 mm, and an outer diameter of 10 mm.
mechanical integrity [3]. Other groups have also
investigated micromachined planar PM generators. One
has demonstrated a self-contained flow-driven generator
that produced 1.1 mW at 30 krpm for a 7.5 mm diameter
Rotor Stator
rotor containing small NdFeB magnets [4]. Another has
reported 14.6 mW output power for an 8 mm diameter
machine at 58 krpm using a NdFeB rotor [5], but in the
corresponding conference presentation, presented late-
Cross-sectional cut
CENTERLINE

breaking results of 5 W of power at 380 krpm.

Rare-earth PMs (e.g. SmCo, NdFeB) are the rotor material
of choice for these devices because the high energy mo Rotor Back Iron
products offer the highest electromechanical power mi
tbi Rotor Magnet
conversion. However, when considering future integration Inner End-Turns tmag Outer End-Turns
of the generator in a high-temperature microengine [1], g
SmCo is preferred (over NdFeB) for its high Curie tend
ttcond
cond
temperature and low remanence loss at elevated tsub
temperatures. hi ho
ri Radial Conductor Stator Substrate
Thus, in furtherance of these efforts, we now report a ro
redesigned, second-generation generator having a similar
Figure 1. Axial-flux permanent-magnet machine
construction, but with optimized machine geometries. This
perspective and cross-sectional schematics.
paper highlights the design considerations and fabrication
 Form Approved
Report Documentation Page OMB No. 0704-0188

Public reporting burden for the collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and
maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information,
including suggestions for reducing this burden, to Washington Headquarters Services, Directorate for Information Operations and Reports, 1215 Jefferson Davis Highway, Suite 1204, Arlington
VA 22202-4302. Respondents should be aware that notwithstanding any other provision of law, no person shall be subject to a penalty for failing to comply with a collection of information if it
does not display a currently valid OMB control number.

1. REPORT DATE 3. DATES COVERED

2. REPORT TYPE
2005 00-00-2005 to 00-00-2005
4. TITLE AND SUBTITLE 5a. CONTRACT NUMBER
Optimization of a Microscale, Axial-Flux, Permanent-Magnet Generator 5b. GRANT NUMBER

5c. PROGRAM ELEMENT NUMBER

6. AUTHOR(S) 5d. PROJECT NUMBER

5e. TASK NUMBER

5f. WORK UNIT NUMBER

7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) 8. PERFORMING ORGANIZATION

REPORT NUMBER
Georgia Institute of Technology,School of Electrical and Computer
Engineering,Atlanta,GA,30332
9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10. SPONSOR/MONITOR’S ACRONYM(S)

11. SPONSOR/MONITOR’S REPORT

NUMBER(S)

12. DISTRIBUTION/AVAILABILITY STATEMENT

Approved for public release; distribution unlimited
13. SUPPLEMENTARY NOTES

14. ABSTRACT

15. SUBJECT TERMS

16. SECURITY CLASSIFICATION OF: 17. LIMITATION OF 18. NUMBER 19a. NAME OF
ABSTRACT OF PAGES RESPONSIBLE PERSON
a. REPORT b. ABSTRACT c. THIS PAGE
4
unclassified unclassified unclassified

Standard Form 298 (Rev. 8-98)

Prescribed by ANSI Std Z39-18
 3. DESIGN IMPROVEMENTS (a) (b)
There are several design improvements in the second- B+ B-
A+ A- C+ B+ A-
generation device aimed at reducing the winding resistances C- A+ C+
B-
by making more effective use of the available volume C-
(between the rotor magnet and stator substrate). First, the
radial conductor segments occupy the full thickness of both
metal layers, as depicted in Figs. 1 and 2. They are thicker,
200 µm rather than ~100 µm, and therefore have lower
resistance. Second, the inter-conductor spacing was reduced
from 130 µm to 50 µm, further increasing the Cu cross-
sectional area for lower resistance. Third, the radial
conductors were shortened by 500 µm relative to the rotor
magnet radial span, permitting optimal flux linkage of
radial leakage flux while minimizing the total coil length.
The new design also features a more effective winding
pattern with an improved end-turn connection scheme, as Figure 2. Stator winding patterns for (a) original, 8-pole, 2-
shown in Fig. 3. The number of “crossovers” and thus vias turn/pole and (b) optimized, 8-pole, 3-turn/pole.
was reduced (from 96 to 3) by using variable-pitch coils
and permitting each end turn segment to occupy only one
layer (either top or bottom). Also, each end-turn segment
width, length, and shape were optimized for minimum
overall resistance.
4. MACHINE OPTIMIZATION
The design goal is to maximize the flux linkage between the (a)
rotor and stator for maximum induced voltage while
minimizing the stator winding resistance for maximum
output power. For a generator of a fixed volume operating
at a constant rotational speed, first-order analysis indicates
that output power scales independently of the number of
poles (p) and number of winding turns/pole (n)—increasing
either has the effect of increasing the output voltage while (b)
simultaneously increasing the winding resistance. However,
when design and fabrication limitations are introduced, the Figure 3. A comparison of the winding diagrams for the (a)
simple first-order approximations do not hold. 2-turn/pole original machine and (b) 3-turn/pole optimized
machine. Solid lines represent layer 1 and dashed lines,
Therefore a parametric optimization was performed to find layer 2. Phase A is darkened for reference.
the optimum values for p and n along with the
corresponding optimal winding pattern geometries for Table 1. Optimization parameters.
maximum output power. Both p and n were varied while
FIXED PARAMETERS
enforcing certain microfabrication constraints for the rotor Parameter Description Value Units
and stator (Table 1). The radial dimensions of the magnet mo Magnet outer radius 5.0 mm
were fixed, as were all axial parameters (e.g. air gap, mi Magnet inner radius 2.5 mm
thicknesses, etc.). A four-step procedure was used. ro Stator radial conductor outer radius 4.75 mm
ri Stator radial conductor inner radius 2.75 mm
1. For each combination of p and n, the inner and outer end- tbi Back iron thickness 500 µm
turn extensions, hi and ho, were varied to find the winding tmag Magnet thickness 500 µm
pattern that yielded the lowest electrical resistance. The tcond Radial conductor thickness 200 µm
tend End turn thickness 80 µm
relative resistance contribution of each winding segment tsub Substrate thickness 1000 µm
(radial conductors, inner end turns, and outer end turns) as g Air-gap 100 µm
well as the total resistance was also calculated using wmin Minimum feature width 50 µm
geometrical relations using a resistivity of ρCu = 1.7 µΩ·cm. Rpe Power electronics equiv. resistance 100 mΩ

2. For each value of p, a 3-D, nonlinear, finite-element VARIABLE PARAMETERS

model (FEM) was used (FEMLAB) to solve for the static Parameter Description Range Units
magnetic B-fields in the machine. The rotor was modeled p # poles 2 - 12 -
n # turns/pole 1-6 -
using an ideal magnetization with a remanence of Br = ±1 T
ho Outer end-turn extension 0 - 2.5 mm
and relative permeability of µr = 1. The FeCoV back iron hi Inner end-turn extension 0 - 2.5 mm
and NiFeMo stator substrate were modeled using 5. FABRICATION
experimentally measured nonlinear material properties. The To confirm an improvement in the overall design,
model employed two magnetic half-poles and enforced optimized stators were built using previously reported [2]
periodic boundary conditions, as shown in Fig. 4. methods. The final device consisted of two electroplated Cu
layers, 80 µm and 60 µm thick, respectively, insulated by
3. Using the optimized winding dimensions and the results
40 µm of SU8, as shown in Fig. 6. The fabricated geometry
from the FEM, the induced voltage waveforms and rms
differs from the modeled geometry (80 µm Cu layers with
voltages were computed. This was accomplished using a
20 µm insulation) due to process variations. The single-
MATLAB script to numerically integrate the B-field to find
phase winding resistance and inductance for this optimized
the flux through an area defined by the shape of the coil.
machine are 160mΩ and 0.31 µH, respectively, higher than
The integration area was then “rotated” in 1° step
the projected values of 100 mΩ and 0.138 µH.
increments to simulate the relative motion of the rotor.
This process was used to determine time rate of change of
flux through the coil, and hence the induced single-phase 2 mm
(a)
open-circuit voltage. (b)
4. The rms value of the open-circuit voltage was then
applied to a matched-load circuit model using a series
equivalent resistance of 100 mΩ to account for losses in the
power electronics. Inner end turns
Fig. 5 shows the matched-load output power, normalized by
the maximum, indicating that 8-pole, 3 turn/pole is the
optimal configuration. The unplotted regions indicate
Outer
geometries that exceed the limitations of the fabrication 500 µm End-turns
design rules, i.e. the conductors are too narrow.
Radial
Conductor

Substrate (d) Outer end turns (c)

Figure 6. (a) Fabricated 3-turn, 8-pole stator, (b) Close-up

on inner end-turns, (c) Close-up of outer end-turns and
electrical connection arm, (d) SEM cross-section of
machines (compare with Figure 1).

Figure 4. FEMLAB finite element model results showing 6. RESULTS

the z-directed B field 100 µm above the stator surface for Using the test stand and approaches described in [2],
the case of p=8. electrical characterizations were performed. The open-
circuit voltage Voc is linear with speed and decays with air
gap, as depicted in Fig. 7. The experimental values are seen
to closely match analytical models.
Maximum (a) (b)
performance

Figure 7. Open-circuit voltages (a) vs rotational speed for

100 µm air gap and (b) vs. air gap at 100,000 rpm for
original and optimized machines. Data points represent
Figure 5. Matched-load three-phase output power,
measurements; lines represent analytical model.
normalized by the maximum, at 305 krpm, indicating
maximum performance for 8-pole, 3-turn/pole stator.
The new, optimized machine exhibits a 26% higher open-
circuit voltage as compared to the original machine. This
increase ideally should be 50% (3 turns/pole rather than 2
turns/pole), assuming equal magnetic properties, magnetic
surface area, and magnetic gap. However, the original
machine possessed a 9% larger magnet surface area (3.2
mm ID, 9.5 mm OD, compared to 5 mm ID, 10 mm OD). In
addition, the magnets were manufactured by two different
vendors, and may not possess equal magnetic properties.
Also, there is some variation in the magnetic gap.
Therefore, a direct comparison of performance is difficult.
For power measurements, the air gap was fixed at 100 µm.
A passive AC/DC converter, comprising a three-phase
∆/wye-connected (1:6 turns ratio) transformer and a three-
phase diode bridge rectifier, was used to provide DC power
to a resistive load [2]. The output power was measured as a
function of load resistance at 100 krpm to find the peak
output power (matched-load condition). The load was then
fixed, and the power was measured as a function of speed.
The load resistance was re-measured at each speed to
ensure accuracy as the resistor heated.
For these experiments, a maximum DC output power of 8
W was achieved at 305 krpm for a matched 37 Ω load, as
shown in Fig. 8. For an active volume of 136 mm3 (OD=10
mm, ID=5 mm, thickness=2.3 mm), this corresponds to a
power density of 59 W/cm3.
The predicted electrical, ηe, and generator, ηg, system
efficiencies [2] under maximum power transfer conditions
(i.e. matched-load) are shown in Fig. 9 as a function of
speed. The electrical system efficiency, which neglects
magnetic eddy current losses in the substrate, approaches
the ideal 50% as the speed increases. This shows how
transformer core losses and diode voltage drop losses in the
AC/DC converter become less significant with increasing
speed. The generator system efficiency is smaller than the
electrical system efficiency due to eddy current losses in the
stator core. This indicates the need for magnetic
laminations in the stator.
Figure 8. Matched-load DC output power for original and
optimized machines. Data points represent measurements;
lines represent analytical model.
Electrical Efficiency
Generator Efficiency
Figure 9. Electrical system efficiency, ηe, and generator
system efficiency, ηg, vs. speed (matched-load conditions).
7. CONCLUSIONS
The optimized machine demonstrated 16 W of mechanical-
to-electrical power conversion and delivery of 8 W of DC
power to a resistive load at a rotational speed of 305 krpm
(59 W/cm3 power density). A maximum power of 1.1 W
was achieved for the first-generation machine at 120 krpm
[2]. At the same speed, the second-generation machine
demonstrates 1.3 W and a corresponding 30% improvement
in power density attributed solely to stator enhancements.
Combining these stator improvements with the higher
rotational speeds (305 krpm) enabled a 7.2x improvement
in total output power. These speeds represent improvement
compared to the highest speeds previously reported [2,3],
but unfortunately, at 305 krpm, the SmCo permanent
magnet suffered catastrophic mechanical failure and
destroyed the stator.
Future work will continue to focus on improving the power
electronics, reducing substrate eddy current losses, and
determining the machine performance at elevated
temperatures. Other work may focus on investigating
NdFeB rotors for potential low-temperature applications.
ACKNOWLEDGMENTS
This work was supported by the United States Army
Research Laboratory Collaborative Technology Alliance
(DAAD19-01-2-0010).
REFERENCES
[1] S. A. Jacobson and A. H. Epstein, “An informal
survey of power MEMS,” ISMME 2003.
[2] S. Das, et al., “Multi-watt electric power from a
microfabricated permanent-magnet generator,” IEEE
MEMS 2005, pp. 287-290.
[3] D. Arnold, et al., “High-speed characterization and
mechanical modeling of microscale, axial-flux,
permanent-magnet generators,” Transducers ‘05, pp.
701-704.
[4] A. S. Holmes, et al., “Axial-flux permanent magnet
machines for micropower generation,” J.
Microelectromech. Syst., vol. 14, no. 1, pp. 54-62,
Feb. 2005.
[5] H. Raisigel, et al., “Magnetic planar micro
generator,” Transducers ‘05, pp. 757-761.